Work

Differentially Private Markov Chain Monte Carlo

Year: 2019

Type: conference-paper

Abstract: Recent developments in differentially private (DP) machine learning and DP Bayesian learning have enabled learning under strong privacy guarantees for the training data subjects. In this paper, we further extend the applicability of DP Bayesian learning by presenting the first general DP Markov chain Monte Carlo (MCMC) algorithm whose privacy-guarantees are not subject to unrealistic assumptions on Markov chain convergence and that is applicable to posterior inference in arbitrary models. Our algorithm is based on a decomposition of the Barker acceptance test that allows evaluating the Rényi DP privacy cost of the accept-reject choice. We further show how to improve the DP guarantee through data subsampling and approximate acceptance tests.

Source: Työväentutkimus Vuosikirja

DOI: 10.48550/arxiv.1901.10275

Authors: Heikkilä, Mikko A., Jälkö, Joonas, Dikmen, Onur, Honkela, Antti

Institutions: University of Helsinki, Helsinki Institute for Information Technology, Aalto University, Halmstad University

Language: English

Topic: Privacy-Preserving Technologies in Data

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funder: Finnish Center for Artificial Intelligence

OpenAlex ID: W2911995276